Computing is still in its infancy, says Keith Collins, CTO for SAS, who believes that analytics and “big compute” are the new paradigm for the foreseeable future. Datanami spoke with Collins about the shift that is underway and what it means for organizations that are experiencing this evolution.
|Keith Collins, Senior Vice President and Chief Technology Officer, SAS
Collins warns that recognizing the shifting winds and taking the right actions is critical for business leaders to get the most of their investments as they transition in this next phase of computing. “We’ve been doing this for less than half a century,” he said, explaining that there is still a lot of realization happening on what can be accomplished as computing evolves. “We’re going to go through many morphs; we’ve learned to run the plumbing and the piping, and now we’re recognizing that just doing that is no longer sufficient a value-add as it used to be.”
Among the first challenges that businesses executives face, says Collins, is grappling with the idea of analytics and recognizing what enterprise can accomplish through their use. “Analytics is nothing more than what the human brain has done all along,” he explains, “making associations, predicting probabilities, trying to optimize. What we’re really seeing is that we’ve come to an age where the level of compute capacity that we have, with the amount of data that we have flowing at us, it requires that we now apply a level of analytics in almost anything that we do. We have to filter out the noise.”
Using a customer example of days past, Collins illustrates the point using the local hardware store owner who would know everyone that came into his store and understand their needs when they walked in. “What we’re really trying to do [through analytics] is that type of things on a larger scale, when it’s beyond any single person’s capacity to keep track of all the data points that are being collected.”
Business leaders face significant challenges as they transition their companies to bring analytics to bear in their organizations, he explained; none more so than the CIO. Collins explains that as the industry has come to this point where data and analytics intersect with big compute power, CIOs need to adjust how they view their role in order to enable their team to be of greater value to the organization in this transition.
The industry has crossed a threshold where IT understands all the moving parts of the business better than almost anybody else, says Collins, leaving it up to the CIO to help stitch together the strategy across the business and help the organization understand how to leverage the data analytics technologies that are becoming increasingly available.
“A great place to start is like all places – how do you skill your teams appropriate to answer this new set of questions,” he explained. “If your questions are no longer about putting the latest desktop down, and no longer about keeping the lights on for running your ERP, but your questions are ‘help me answer the things I don’t know,’ then you need to skill your teams for that.”
Collins says that this isn’t done by having better project or alliance managers, but instead it’s about getting them skilled in the techniques of answering the questions around analytics. “What does it mean to do predictive modeling; what does it mean to do optimization; what does it mean to do forecasting?” Having an organization skilled in knowing how to answer these questions helps people understand how to organize the data, explains Collins.
“If we’re going to democratize data, then data basically need to be free in your organization, not locked up,” cautioned Collins, clearly aware of the can of worms this subject raises. “That’s counter to the premise of ‘Mr. CIO, you’re responsible for our security and that governance of data,’ – the natural answer is, ‘well, we’re just going to lock it up.’” That’s exactly the opposite of what needs to happen, he says.
Does that mean knocking down the silos and dumping everything into a single unified database? Not so fast, says Collins, who explains that the silos help give appropriate structure and ownership to the data relative to the answers that it needs to deliver. “If we don’t fight against the silos, but instead add value to what’s in the silos, then maybe they start to get convergence,” he says.
Collins argues that if you take away the autonomy of each units’ data, ultimately what you have done is slowed the business units down for their specific business objectives and effectively disenfranchised them from the data that they’re responsible for. Using the fraud team as an example, Collins says that it’s counter-productive to take the data away from them in order to reorganize it with data across other parts of the organization. He likens this to taking the data away from them and locking it up.
“I’m a huge fan of techniques like data federation and master data management to govern across these data sources and puts data in the right framework for its usage, and learn how to leverage it across the infrastructure as opposed to this mantra of ‘I’m just going to have one version of the truth,’” he says. “When people say, we’re just going to put all the data in one place and everybody is going to get it from here, there is a fair amount of fallacy to that when you actually need to organize the data for the outcome of what you’re trying to do.”
Outside of ensuring proper skills on the IT team, and solid data governance plan, the third leg of the stool, says Collins, is providing proper access to the analytics tools that each team needs. This includes the tools that allow the hypothesis, the visualization and the analytics that happen around data. To help facilitate this, Collins says he is a fan of “shadow IT” teams.
“The CIO needs to be mature enough to know that embracing shadow IT is the best thing for an organization to bring collaboration across the business units – to bring those skills of analytics that are in the risk group, the marketing group, the fraud group together,” he says.
Collins says that in his own organization (SAS), they have a core IT team, with small IT teams spread in departments around the organization to help make sure that each team has exactly the data that they need, the shape that they need it in, with the tools that they need to get to answers.
Ultimately, Collins says that while the “big data” is getting all the attention, the shift that is happening is really about big compute and high performance analytics, and CIOs will profit from understanding how to adapt in this new environment. “That’s really the change that is happening – some people will have more data; some people will have data that comes in bigger quantities; some people will have more variety. These are all true – but the big difference is now there is a whole new class of compute power and analytics that apply to help you be more agile with your business processes.”
Hadoop Distros Orbit Around Solr
Can Cloud Bridge the Big Data Talent Gap?
SAS Announces New High-Performance Analytics Products